VIDEO: Huaiying Zhang on Liquid-Liquid Phase Separation in Telomere Maintenance of ALT Cancer

Huaiying Zhang (02:21):
So when I transitioned into biology for my postdoc, I was fascinated to find out that this phenomena is actually used in a cell. On the left you can see Cliff’s pioneer work looking at the liquid behavior of P granules and look at how they drip, how they fuse. This led to the whole field of phase separation as an organizer for biochemistry. On the right, you can see the beautiful droplets reconstituted by the Rosen lab, which reviewed the multivalency as the driving force for phase separation. Next, please.

Huaiying Zhang (02:59):
You can imagine how excited I was when I was joining the Summer Institute as a postdoc with Amy Gladfelter to work on protein phase separation, it’s just all coming around. Next, please.

Huaiying Zhang (03:14):
So the protein we were working on is a polyQ containing protein, as Mark was referring to. It is a RNA-binding protein, it has it’s RNA recognition motif at the end terminus, and it has this polyQ stretched in the middle. Next, please.

Huaiying Zhang (03:33):
This polyQ’s domain is highly disordered as we can see in this prediction here. Next, please.

Huaiying Zhang (03:43):
So we purified this protein, we see nice droplet formation. Interesting, the formation of the droplets depends on the polyQ domain, however, the polyQ domain is not sufficient to drive phase separation of this protein alone. Next, please.

Huaiying Zhang (03:58):
And then we look at how the RNA binding affects the phase behavior at physiological condition, meaning at 150mM of salt, at nanomolar protein concentration. The protein alone does not phase separate. However, when we add the RNA that can be specifically interacting with the protein, we see droplet formation. We don’t see the droplet formation when we add non-specific RNA, meaning total RNA to normalize the charge. This indicates that it is a specific interaction between the RNA and the protein that can shift the phase boundary of the protein. Next, please.

Huaiying Zhang (04:35):
And then I went to Cliff’s lab where we used microrheology to look at material property of the droplets. You can see here in the droplets we using red for the beads that we embedded in the droplets and then the green is the protein that is labeled and the RNA is not labeled in this case. Then we tracked the movement of those beads and then we plotted their displacement over time. Next, please.

Huaiying Zhang (05:02):
As in MSD versus time curve. Next, please.

Huaiying Zhang (05:07):
And then you can see as RNA concentration increases from the red to the green, you can see that the beads moved slower, meaning that the droplets become more viscous indeed. Next, please.

Huaiying Zhang (05:21):
When we get the viscosity from the movement of the beads, we can see that as the RNA concentration increased, the viscosity of the droplets increases. So this is a very nice tool which will allow us to determine what driving and what are the domains of the protein and what are the interactions that can drive phase separation, and how the material property is affected by different components in the droplets. Next, please.

Huaiying Zhang (05:52):
So when I started my lab at Carnegie Mellon University, I really wanted to look into, what is the relevance of phase separation and how does that to do with cell function and with diseases. In particular, we’re interested in looking into how does phase separation can be linked to cancer. How do cancer cells use this process? Or what went wrong for several liquid phase separation processes. Next, please.

Huaiying Zhang (06:18):
I’m going to share with you our work on particular cancer cells that use a specific way to maintain their telomere length. All cancer cells have to actively maintain their telomere length, and this is because the telomere will be shortened with every cell division because the replication machinery cannot duplicate all the way to the end of the chromosome. Therefore, every time the cell divides, the telomere gets shortened. And when telomeres become too short, it triggers a cell cycle arrest. Therefore, for all cancer cells to keep dividing, they have to actively maintain their telomere length. Next, please.

Huaiying Zhang (06:59):
Majority of cells reactivate telomerase, which is a reverse peptidase that use RNA as a template to add back the lost telomere DNA. However, there’s a significant amount of cancer cells actually do not reactivate telomerase. What they use is called a alternative lengthening telomere pathway, we call it ALT pathway. This is a pathway that is based on homology directed to DNA repair, meaning that one telomere uses another telomere as a template to repair and the details of molecular mechanism are still under investigation. As you can see from this chromosome spread, a feature of all those cancer cells is that their telomere length is very heterogeneous. You can see the pink, which stains for the telomere, is very heterogeneous in those cancer cells, compared to a on-ALT cancer cell, for example.

Huaiying Zhang (07:54):
Interestingly, for cancer cells that are telomerase positive… Next, please. If they are subject to telomerase inhibitor, some of them will die, but some of them actually will start to adopt this ALT pathway for telomere maintenance. And this becomes an increasing problem for cancer therapy using telomerase inhibitors. Next, please.

Huaiying Zhang (08:19):
So in addition to the heterogeneity in telomere length, there are many other unique features of these ALT cancer cells that are used for diagnosis. Because unlike telomerase-positive cancer cells, ALT cancer cells do not have a molecule specifically that can be used as a unique marker. Therefore, it use a combination of several phenotypes. One of them is the relocation of the PML body to the telomeres. PML bodies are membraneless organelles that exist in the nucleus of normal cells and the number and size change depending on the cell cycle, depending on the status of the cell, such as stress conditions. However, only in those cancer cells, the PML bodies are find to localize to the telomere. As you can see here, it’s showing you the image of some cancer tissue, the PML bodies are stained by the PM protein, localized perfectly with the telomere DNA. Interestingly, the telomeres actually cluster within APBs, meaning that they show as big puncta as shown here in this red staining of the telomere DNA. And this clustering is thought to provide template, to bring telomeres together so one can use another as a repair template for homology-directed DNA repair. Those are used in diagnosis, then they got people asking “Can we use this phenotypes for cancer therapy then?” And this is hindered by our… Next, please.

Huaiying Zhang (09:52):
There’s a lack of understanding, how actually do APBs form? Meaning, how do those PML bodies are relocalized to the telomere? Are they actually nucleated from the telomeres? Or, they actually are the existing PML body, but somehow relocalized to the telomeres. And then how do they contribute to the ALT cancer pathway? Next, please.

Huaiying Zhang (10:18):
So the the Roger Greenberg Lab at Penn, they developed the assay by introducing DNA damage on the telomere, they find out that this damage response on the telomere can lead to APB formation, telomere clustering with the APB, and other phenotypes of the ALT pathway including telomere elongation. So, when I was collaborating with them at postdoc in Michael Lampson’s lab, and I looked at these movies of telomere clustering and I thought “Oh, these just look like droplets fusing.” So, showing here on the bottom left, you can see this process of the reduction in aspect ratio where the APBs actually round up, and if we FRAP them we will see the dynamic exchange of components. Next, please.

Huaiying Zhang (11:05):
This led me to hypothesize… Next, please.

Huaiying Zhang (11:09):
That those might be actually APB bodies formed as a response of DNA damage on those telomeres. And then, APB as a condensate, with liquid behavior… Next, please.

Huaiying Zhang (11:22):
That will allow them to fuse and this fusion process then will allow telomere cluster within the APB. Next, please.

Huaiying Zhang (11:32):
Moreover, this concentration process can provide a unique opportunity for the APBs, actually, to selectively enrich DNA repair factors on the telomeres and to allow DNA repair to happen. What’s unique for human telomeres is that [inaudible 00:11:51], so that the telomere has specific structure property to prevent the cells from recognizing as a damaged [inaudible 00:12:02]. Therefore, actually there are active mechanisms in place to prevent DNA repair factors to be localized to the telomeres. Therefore, the formation of APB maybe can provide this unique chemical environment that actually allows the DNA repair factors to be enriched. Next, please.

Huaiying Zhang (12:21):
And then after clustering telomeres together and the enrichment DNA repair factors, the APB maybe then can be a place where allows telomere elongation to happen. Next, please.

Huaiying Zhang (12:34):
So to test this hypothesis, we ask ourselves, “How do we test this?” Because DNA damage response is a really complex process. In addition to APB formation, there are many other things going on. For example, DNA damage foci deformation and that a lot of proteins are not part of the APB process. And then how do we decouple the contribution of those from the APB then? Our approach was to try to induce de novo APBs without DNA damage. That way, we can specifically decouple this function from APB formation, from other aspects of DNA damage response. Next, please.

Huaiying Zhang (13:15):
So, one tool that will allow us to do so is optogenetics, which uses light to control genetically coded light-sensitive proteins. Next, please.

Huaiying Zhang (13:25):
These tools are widely used in a lot of cellular processes, particularly those dynamic cellular processes where spatial-temporal organization is essential, such as phase separation. Because by using the laser, we can have really good spatial and temporal control of where we want things to happen. Then today I’m going to share with you some of the tools we developed. Instead of using light-sensitive proteins, we create small molecules that are light sensitive which provide some unique properties that allow us to control phase separation in cells. Next, please.

Huaiying Zhang (14:02):
So those tools are based on a modular design called photocaged chemical dimerizer where we choose two small molecules that can each interact with a protein. Next, please.

Huaiying Zhang (14:13):
The red pair is a small molecule called TMP that can interact with bacterial DHFR and this interaction is non-covalent. Next, please.

Huaiying Zhang (14:24):
On the right, the green is a HaloTag system where the Halo ligand binds with the Halo enzyme, this is a covalent binding. Next, please.

Huaiying Zhang (14:33):
And then we put a linker between those two small molecules, the TMP and the Halo, that becomes the dimerizer. Next, please.

Huaiying Zhang (14:43):
To make this photosensitive, what we did was we put a photocage on the TMP so that this dimerization is prevented unless we use a light to get rid of the photocage. Next, please.

Huaiying Zhang (14:57):
And then the photocage we use is NVOC. Next, please.

Huaiying Zhang (15:01):
And this is the structure of the dimerizer where we call NTH, for NVOC, TMP, Halo. And this system is unique compared to other optogenetic systems. It’s based on heterodimerization, meaning that we can dimerize two different proteins together. And also, we particularly chose the HaloTag system to be covalent and the TMP to be non-covalent, so that we can use actually the Halo to anchor dimerizers to a subcellular localization. Therefore, it would allow us to recruit proteins specifically to a subcellular localization. Next, please.

Huaiying Zhang (15:39):
I’m going to show you an example of how we used this tool to recruit a protein. In this case, just another protein of mCherry to centromeres, which is labeled in green where we fused the Halo to a centromere protein. As you see here in the green is the centromere and then the red is the mCherry which is diffusively localized in the cell. If we used a wide-field, just whole field of UV to activate the dimerizer, to get rid of that photocage and to turn on the dimerization. Next, please.

Huaiying Zhang (16:15):
You will see the mCherry now is recruited to where the centromeres are. Next, please.

Huaiying Zhang (16:21):
Now we can also achieve spatial control, in that we only want to recruit to one centromere. The advantage of doing that is we can use other centromeres as the internal control to look at the behavior of that specific centromere. Next, please.

Huaiying Zhang (16:39):
So showing here is, again, the centromere and in this case we used a 405 laser to specifically target to one centromere. Next, please.

Huaiying Zhang (16:47):
As you can see, before the laser, and we don’t see a centromere there. Next, please.

Huaiying Zhang (16:55):
Once we turn on the laser and now you can see recruitment, in a couple seconds, and then the mCherry is recruited to only one of the centromeres. Next, please.

Huaiying Zhang (17:07):
We found this toolbox to include other functionality by shifting things around. This is advantage, again, of this modular design. You’ve seen the NTH, which is a caged version. Next.

Huaiying Zhang (17:25):
If we move the NVOC in the middle, now it become an uncaged, cleavable version. Meaning that you can add the dimerizer proteins will be recruited to where the subcellular locations you want, but then you can use light to reverse the dimerization and then release protein from that localization and then follow its functionality. Next, please.

Huaiying Zhang (17:47):
We also created a version where we use a different photocage. This use a different wavelength to uncage the dimerizer. Next, please.

Huaiying Zhang (17:57):
And this allows us to combine them together because they are orthogonal. The NVOC is activated at 405 laser, and then the Coumarin is activated at 444 laser. So now we have a caged and a cleavable dimerizer where we can use 405… No, where we can use actually the Coumarin on the cage. So we can use 444 to control the recruitment and then we can use the 405 to cleave the molecule and then release the protein. So we’ve created a library of these dimerizers and then another unique property of the dimerizers is you can’t have the same engineered cell background but by switching dimerizers, allows you have different functionality. For example, you can use the uncaged dimerizers, just with the TMP and Halo, you can add it to your population with cells, do biochemistry, etcetera. And also, once the cage is cleaved and then the dimerization is persist through a long time, so you can use long term imaging without using the light to keep the dimerization on. So we used those two subcellular processes including organelle control, kinetochore function… Next, please.

Huaiying Zhang (19:21):
In this case, in my lab, we want to adopt these tools to study APB function. To induce APB without DNA damage. So, what we want to do is to recruit a protein to the telomeres and then this protein will lead to APB formation without DNA damage. Then we asked “What kind of proteins do we recruit, then?” Next, please.

Huaiying Zhang (19:47):
This requires us to ask “What could be the driving force for APB condensation?” Next, please.

Huaiying Zhang (19:54):
And then we went to literature, we find another unique feature for those cancer cells is somehow their telomere proteins are uniquely modified by Small Ubiquitin-like Modifier. And this sumoylation process… Next, please.

Huaiying Zhang (20:10):
Is required for the APB formation. Next, please.

Huaiying Zhang (20:16):
Also, we look into the components in the APB condensate. And what we find out is a lot of the components either have sumoylation sites or they contain SIM, which is a small motif that can non-specifically interact with SUMO. Next, please.

Huaiying Zhang (20:34):
And the Michael Rosen lab have already shown that the multivalent interaction, this non-covalent interaction, between SUMO-SIM can lead to phase separation. Therefore, our hypothesis became… Next, please.

Huaiying Zhang (20:49):
That the APB formation is fundamentally triggered by a telomere sumoylation. This SUMO-SIM interaction then drives APB condensation. Next, please.

Huaiying Zhang (21:05):
In agreement with this, we look at whether actually SUMO is enriched on the telomere after we induce DNA damage on the telomere which leads to APB formation. So in this case, we used the Roger Greenberg assay where we used the nuclease Fok1 to induce DNA damage on the telomere. And here in this case, the Fok1 is tethered, it’s fused to the telomere binding protein TRF1, so what you are seeing in Fok1 is where the telomeres are. And you can see the SUMO, which is the green, localized perfectly with the Fok1 telomere signal, meaning that we see enrichment of SUMO on the telomere after DNA damage. Next, please.

Huaiying Zhang (21:46):
This is not seen when we used a Fok1 mutant, that is enzymatically dead, meaning that it is after DNA damage response the SUMO is enriched on the telomere. Next, please.

Huaiying Zhang (22:00):
And just showing you the quantification where you can look at the percent of telomeres with SUMO foci. We can also look at intensity, both with the significant difference between those two cases. Again, meaning that the telomeres are sumoylated after DNA damage response. Next, please.

Huaiying Zhang (22:22):
So, this leads to the hypothesis where the DNA damage response that triggers APB formation is the sumoylation process. So after telomere proteins are sumoylated, then it can recruit proteins that contain SUMO-SIM, which may include PML. Which has three sumoylation sides and the one SIM. So after the enrichment, SUMO-SIM on the telomere, then the local enrichment of SUMO-SIM leads to the multivalent interaction because SUMO-SIM then leads to phase separation and the formation of APB. Next, please.

Huaiying Zhang (23:03):
So we want to use our optogenetic tools to mimic this sumoylation process without DNA damage. And then what we want to do instead of recruiting SUMO, we want to recruit SIM instead because we don’t want to overexpress SUMO, which is involved in lots of processes. And so, the idea is if we recruit SIM to the telomere, then the SIM will then bring the endogenous SUMO there to mimic the sumoylation process and then recruit other proteins contacting SUMO-SIM then leads to phase separation. Next please.

Huaiying Zhang (23:40):
And to do that we fuse the SIM to the eDHFR and mCherry for the visualization. And then we fuse the Halo to the telomere binding protein TRF1 so that the dimerizer is anchored on the telomere. And then we first use immunofluorescence to see if we recruit SIM can we actually enrich SUMO like we hypothesized. Indeed, as shown here in this, you can see perfect colocalization between SIM and the telomere and the SUMO foci. Next, please.

Huaiying Zhang (24:18):
And this is with wild-type SIM. However, if we use a SIM mutant that cannot interact with SUMO we don’t see such enrichment. Meaning that it is the specific interaction between SUMO and SIM that is enriching SUMO at the telomere after dimerization. Next, please.

Huaiying Zhang (24:37):
So then we use live imaging to visualize this process, what happens once we recruit SIM to the telomeres. And this is before dimerization and the SIM is diffusive on the left and then the telomere is in the middle which is in the green in the merged image. And when we turn on the dimerization… Next, please.

Huaiying Zhang (25:01):
You can you the foci start to round up, and then they bump into each other and they fuse and you see here in that square that I highlighted for you. So this kind of indicates that there is phase separation going on and there’s also droplets fusing. Next, please.

Huaiying Zhang (25:22):
Okay, good. This the snapshot of some and you can see the SIM is recruited initially and then starts to round up. And you can see actually Halo, TRF1 channel actually also starts to round up as well. And this is because the TRF1 that after we added dimerizers, the TRF1 is actually physically linked to the SIM. So the SIM undergoes phase separation. And the TRF1 also undergoes phase separation with SIM. So it behaves just like the SIM. Next, please.

Huaiying Zhang (26:01):
And then when we quantify the intensity of foci per foci. The integrate intensity, which indicates the size and also the intensity, the brightness of the foci. And you can see that SIM, which is in magenta, start to increase with the recruitment and also because of phase separation. And you can see the TRF1 which is the green, also starts to increase as SIM is recruited. As I said, this is because it’s linked to the SIM, so when SIM undergoes phase separation it starts to enrich extra, other than the ones that bind to the telomere DNA. But also this is because telomeres start to cluster so therefore they become bigger. As you can see here on the right, where a quantified number of foci, over time you can see they start to decrease after SIM is recruited, after the phase separation takes place. This is because of telomere clustering. Next, please.

Huaiying Zhang (27:01):
I’m showing you a event of this fusion event, which is also telomere clustering because we are looking at the telomeres. And you can see the rounding and the reduction of aspect ratio to one and also dynamic exchange of components by using FRAP, agreeing that this is a liquid condensate that we are forming. Next, please.

Huaiying Zhang (27:24):
And then we look at what happened when we recruit the SIM that cannot interact with SUMO. I told you that with immunofluorescence, we see when we recruit SIM, we don’t see SUMO enrichment. So here by using the same strategy, you can see SIM is on the left and in the middle is the telomere and the telomere is in green on the merged image on the right. Next, please.

Huaiying Zhang (27:51):
When we turn on the dimerization, you can see the dimerizer still works, meaning that the SIM mutant is still recruited to the telomere. However, you don’t see the rounding and the clustering that we see when we recruit SIM. Next, please.

Huaiying Zhang (28:04):
You can see from these images as well. Next, please.

Huaiying Zhang (28:13):
And when we quantify you can see SIM, that’s the magenta, the integrated intensity on the left, the SIM same start to increase over time, but it doesn’t increase further. And if you look at the telomeres, they also behave differently in that the intensity is not increasing anymore. And in fact, they start to decrease because of bleaching. And on the right, you can see the quantification of the number of foci, you don’t see decrease in the number of telomere which is green in this case after we recruit the mutant. So this agrees with the hypothesis where the SIM mutant does not lead to enriching SUMO, does not lead to phase separation. Next, please.

Huaiying Zhang (28:58):
And then we want to see if we can reverse this condensation. So, if the cell used sumoylation to trigger phase separation, then sumoylation is a reverse process. Then for the cells to reverse telomere clustering, they can simply reverse sumoylation. Can we mimic this process then? So what we did then we integrated cells with the dimerizer, then allowed the telomeres to cluster, allowed phase separation to happen and that is the cell that you see here we put on the microscope. And then we start to reverse the dimerization as I play the movie. Next, please.

Huaiying Zhang (29:32):
You can see as we reverse the dimerization of SIM, first it is released from the telomere. But also you can see that the telomere start to become more dilute. Next, please.

Huaiying Zhang (29:48):
As we show you two cases how that happened, is the cases with the dots start to dissociate. Meaning if you look at that bright spot, it becomes two spots. And then on the right showing you that sometimes one dot simply become more diffusive. Meaning that it’s like single proteins are leaving that dot. Next, please.

Huaiying Zhang (30:12):
And then if we look at the quantification again, look at the intensity of the foci over time. And you can see both channels start to reduce and the SIM is because it’s released from the telomere. But the telomere also is reduced, the intensity is reduced because the reverse of phase separation and also reversal of the constraint. If you look at the number, the number of SIM in the magenta, which decreases. But the number of telomeres, they actually start to increase because the reversal of telomere clustering. Next, please.

Huaiying Zhang (30:47):
So, this agrees with hypothesis where sumoylation as the DNA damage response on the damaged or stressed telomere, then can lead to the APB formation. And then the fusion of those APBs, of those nuclear APBs, can lead to telomere clustering. Next, please.

Huaiying Zhang (31:08):
Then we next asked if this phase separation process can actually create an environment enriched in DNA repair factors. Next, please.

Huaiying Zhang (31:22):
So, we hypothesized that the driving force for the APB formation is SUMO-SIM interaction. Therefore, for proteins contacting SUMO-SIM, they can be enriched to this APB condensate. However, we also have a reason to believe this might be more tightly regulated, because we know that, for example, sumoylation has crosstalk with phosphorylation which is another response after DNA damage response. Next, please.

Huaiying Zhang (31:51):
So, to see which one is true and then we just use immunofluorescence to detect those known APB components, and see whether they are recruited to the de novo APBs. Next, please. Next, please. Thank you.

Huaiying Zhang (32:07):
So, here I’m showing you the telomere DNA FISh on the left, and we use… First look at the PML protein, which is the signature component of PML body. And this is just endogenous cells without recruiting anything. And you can see that the PML protein, we can see some colocalization of the PML protein and the telomere. And those are, by definition, APBs. And those are endogenous APBs that exist in those cancer cells. Next, please.

Huaiying Zhang (32:48):
And when we recruit SIM, you will see now there are more white dots, meaning more colocalization of PML on the telomere. Meaning more number of APBs. Next, please.

Huaiying Zhang (33:02):
But this is not the case when we recruit the same mutant that cannot interact with SUMO. Next, please.

Huaiying Zhang (33:10):
And those are quantifications showing the difference in either the number and intensity of the APBs in those cases, where we recruit SIM and we see a dramatic increase in the number and intensity of the APBs. Next, please.

Huaiying Zhang (33:27):
Well, we don’t see the enrichment of DNA repair factors to the induced APBs. And we’ve tested couple including p53bp1, and other repair factors showing which is the cofactor used for the polymerase in those cancer cells. And you can see with the positive control where we use the Fok1 nuclease to induce DNA damage. And you can see PCNA is enriched on the telomere as shown here on the merged image and the number of white foci. Next, please.

Huaiying Zhang (34:03):
And when we recruit SIM, we don’t see such enrichment. Next, please.

Huaiying Zhang (34:09):
The same as for SIM mutants. So we don’t see a lot of PCNA localization to the telomeres after we recruit them. Next, please.

Huaiying Zhang (34:19):
Interestingly, after DNA damage, we know that PCNA is modified by SUMO. And then if we artificially just fuse PCNA to SUMO, in this case to SUMO1. We’ll see that actually after SIM recruitment, we see PCNA now, when it’s fused to SUMO, is highly enriched on the telomere as you can see the colocalized white dots on the right. Next, please.

Huaiying Zhang (34:49):
And this is another thing where we just recruit the same mutant. Next, please.

Huaiying Zhang (34:54):
You can see the difference. When we quantify the number of colocalized foci or the intensity of the colocalized foci on the right. Next, please.

Huaiying Zhang (35:05):
So, this suggests that the induced condensates are indeed APBs because contain the PML protein. However, the other DNA repair factors are not enriched. But if we artificially allow them to be modified by SUMO, then they can be enriched in the condensate. This indicates that this enrichment of the DNA factors, meaning that the composition of this condensate is under tight regulation of DNA damage response. Next, please.

Huaiying Zhang (35:44):
So, if a model of this kind of two function of the APB is true, that indicates that the first function should be decoupled from the second function. Meaning that the first function only relies on the liquid property of the APB condensate. And then the second one depends on its chemical composition. So to test this hypothesis, we want to induce condensates with a different chemistry other than APBs on the telomeres. So we chose/test a couple, I’ll show you one, where we used a disordered region RGG from a LAF-1 protein, which is known to phase separate. Next, please.

Huaiying Zhang (36:26):
And this is showing you here the images before the dimerization. On the left is the RGG and in the middle is telomere. Next, please.

Huaiying Zhang (36:35):
We turn on the dimerization, you will see RGG is recruited, but similar to when we recruit SIM, you can see now RGG start to phase separate because the foci start to round up. Next, please.

Huaiying Zhang (36:50):
As you can see in the images and then again you can quantify the intensity over time of those foci. You see increase in the intensity in both because of the condensation and the clustering. And then the number of foci and changes over time, in both channel after SIM is recruited after the RGG is included because of the clustering. Next, please.

Huaiying Zhang (37:16):
And so this leads to the conclusion, where we show that the sumoylation is the signal that the cell use to nucleate APBs after damage response on the telomere. It’s likely maybe the signal the cells use to dissolve APBs as well. The importance of dissolving APBs is the cells kind of temporarily hold the telomeres together. However, they don’t want to hold them for too long, at least they need to dissolve them before the next cell cycle so that they can have accurate cell division. So we think this reversible sumoylation process is the cue the cells use to nucleate APBs particularly on the telomere after triggering a DNA damage response. And then we’ve also shown that the liquid property of the APBs is sufficient to drive to telomere clustering and however, the recruitment of DNA Repair factors is regulated by other DNA damage signaling. Next, please.

Huaiying Zhang (38:24):
So with that, I’d like to thank the lab. Rongwei and Meng contributed to this work. I also want to acknowledge this work is initiated when I was postdoc at Penn with Lampson. And it was a collaboration with him, it was his lab who created the chemicals for us, and with the Greenberg lab for the ALT cancer. And we are recently funded to study how phase separation organizes the genome of the ALT cancer by the 4D Nucleome consortium and we are looking for postdocs. If you are interested to see, what is the consequences of phase separation and using tools to actually demonstrate those functions and please contact me. With that, I want to thank Mark and Jill for organizing this event and thank you for having me and be more than happy to take any questions that you may have.

Mark Murcko (39:16):
Thank you so much. Thank you wonderful, wonderful lecture.

Huaiying Zhang (39:21):
Thank you.

Mark Murcko (39:21):
You covered so much so quickly and clearly. It was really beautiful. This is beautiful work.

Huaiying Zhang (39:26):
Thank you. Good to hear.

Mark Murcko (39:26):
Yeah.

Huaiying Zhang (39:26):
Encouraging words.

Mark Murcko (39:29):
Yeah. And you were also very gentle about the next slide, please. That all worked out really well. So I think there’s a question from Will Chen. Will, if you want to unmute and ask your question, that’d be great.

Jill (39:49):
Hold on, I’m being a little slow, but I’m going to unmute you Will.

Mark Murcko (39:52):
Good.

Jill (39:58):
All right. There you go.

Will Chen (39:59):
Thank you. Hey, really nice talk. Thank you.

Huaiying Zhang (40:02):
Thank you.

Will Chen (40:05):
So for this concept of upregulating the telomere mechanism… I guess it must be for cancer survival or proliferation. Are there any specific tissue types or genomic biomarkers that tell us… Or is it a pan-cancer mechanism? Is there a specific cancer that depends on this?

Huaiying Zhang (40:29):
So they don’t have a specific tissue, but they are very enriched in soft tissue like nervous system and also some find in the bone. For example commonly used U2OS, bone cancer, and that is ALT positive. And sometimes you will find they have a subpopulation in some of the cancer cells that use ALT, they coexist with cancer cells that use actually telomerase, in the same tissue.

Will Chen (41:05):
Thank you.

Mark Murcko (41:08):
We have a question from… I’m not sure who exactly it is. But Biplab KC. It’s a good question, actually. If Jill could unmute them.

Biplab KC (41:22):
Hello, I am a student from Japan, so it’s midnight here. So I’m sorry, I’m not on my video. My question is…

Huaiying Zhang (41:30):
Thank you for tuning in even though it’s so late. I’m flattered.

Biplab KC (41:36):
Actually I’m doing research on synthetic systems. So I’m quite interested in the biological system of LLPS system. So finally, the one point that I am really interested in is that you dissect the two properties. One is a liquid properties and one is the functional properties of the system. And finally, you said that even if you have the liquid property that allow the clustering of the telomerase, the elongation part of the telomeres itself, you are not able to find, right. Because I think it might need some other proteins, or other factors from the DNA damage response. So I was wondering, what is your next plan then? I mean, even if you can then open the telomerase elongation, finally, is your main target, I think. So what is your next plan? How are you going to tackle this problem? Because the proteins that might be responsible, there may be many proteins, right?

Huaiying Zhang (42:36):
So your question is what proteins are recruited or what… I didn’t get clearly what your…

Biplab KC (42:46):
Telomerase elongation, maybe… I’m not quite familiar in the field, but maybe you need many proteins to be enriched in the liquid phase…

Huaiying Zhang (42:55):
Yes.

Biplab KC (42:55):
… For the telomerase elongation, right?

Huaiying Zhang (42:57):
Yes.

Biplab KC (42:57):
But you cannot find. I mean, you did [inaudible 00:43:01] PCNA using the fusion protein for all of the proteins it may be not feasible practically, I think.

Huaiying Zhang (43:10):
That’s a good question. So there’s recent work… Not from my lab but from other labs showing that one key… PCNA is downstream, right. So it was just showing that can be enriched or not. It’s not going to be trigger the DNA damage response. And I think the recent work showing that there’s a helicase called BLM, which has a SIM domain that is required for that protein to be recruited to APB. Once that protein is recruited, then it can trigger lots of downstream, and then unleash the DNA repair.

Biplab KC (43:48):
Okay. So that means there is some checkpoint. If your [inaudible 00:43:52] was single or some proteins, only then it will downstream for other signaling to recruit other proteins.

Huaiying Zhang (43:59):
Yes. So there’s a series of signaling happening after APB formation even.

Mark Murcko (44:03):
Great. Okay. Let’s go on to… Bede had a question.

Biplab KC (44:09):
Thank you very much.

Mark Murcko (44:11):
Thank you. Bede you’re up.

Bede Portz (44:14):
Great talk. Can you hear me?

Huaiying Zhang (44:17):
Yes.

Bede Portz (44:17):
Yeah. I miss you at Penn. So I might have missed this in the talk. But at the end you you did this last one experiment where you swapped in the last one IDR, which has RGG domains and those RGG domains can can interact with RNA. Do you do you know if generally, there’s an RNA component to the maintenance of these APB bodies and is there transcription occurring locally?

Huaiying Zhang (44:42):
That’s a good question. There is, but we don’t know what is the relationship yet. So there is a telomere RNA that’s being transcribed. In those cancer cells that is upregulated. And if you lock down the RNA you will see phenotype. But how the RNA is involved in this process is not super clear yet. There’s one study showing that the RNA can form a loop to help with DNA damage response. But also there are other hints that the present RNA can help maintain a particular chromatin status of the telomere so that the DNA damage response can happen. Which one it is, or that directly links to APB formation, that is not very clear. But definitely they do play a role by studies showing that if you lock down. They are upregulated, there is upregulation of transcription.

Bede Portz (45:58):
Thanks.

Mark Murcko (46:01):
That’s great. So let’s go to Anirban. That’s a really interesting question about heterogeneity. If you could unmute Anirban.

Anirban Roychowdhury (46:09):
Can you hear me?

Huaiying Zhang (46:15):
Yes.

Anirban Roychowdhury (46:17):
Yeah. Thank you very much for the insightful talk. I have two quick questions. The heterogeneity in the first slide you showed, that there is a heterogeneity in the formation of these telomere bodies. So how in the cells in which it is not forming, does there condensation is not happening? Or what is the phenomena regarding that? And the second question is, there is a G-quadruplex involvement in telomeres maintenance. So how does that affect in the LLPS phenomena?

Huaiying Zhang (46:52):
Good question. The first one you’re asking the heterogeneity in the telomere lengths, right.

Anirban Roychowdhury (47:00):
Right. Right.

Huaiying Zhang (47:02):
So that is an interesting question. We don’t know. We think the reason is the heterogeneity arises from this kind of, not well controlled homologous recombination. Compared to telomerase, where it’s really well controlled because it’s used stem cells. And so the heterogeneity for how do cells deal with that. You have extremely long telomeres, you have extremely short telomeres. And we know that when telomeres become very short, they are going to trigger some response. One of them is actually response to trigger transcription of the telomerase. So we don’t know, actually the shorter ones particularly trigger APB information or not. So when we look at endogenous APB, right. They are not localized to all the telomeres. Looks like only some of them have it. But are those the ones that actually, very short, that trigger the damage response? Or are the ones that happen to be having a lot of sumoylation happening, so then you trigger formation? Those we actually don’t know.

Anirban Roychowdhury (48:18):
Okay. Okay.

Huaiying Zhang (48:18):
Then for the second question is the G-quadruplex. Yes. So the understanding is the telomere in those cancer cells compared to normal cancer cells, they are in some sort of stress condition. Meaning that their telomeres are not well protected. And forming those loops and then they are lacking some protein. But in this kind of stress condition is when DNA damage response is allowed to happen. But in this kind of stress condition, then you might have more G-quadruplex formation, but how does that G-quadruplex formation? Either it’s G-quadruplex of the DNA itself, or actually, there is also reason to believe that the G-quadruplex can form between DNA and the RNA and the RNA itself, then how do cells deal with that? And the the kind of current thought is because the presence of the G-quadruplex, therefore, those telomeres, they have more trouble replicating, and that leads to replication stress, and that replication stress then leads to DNA response and then it become a loop. Allow those cancer cells to survive under those stress conditions, because the stress allows them to survive. You what I’m talking about?

Anirban Roychowdhury (49:39):
Right. It’s a feedback loop. It’s a feedback.

Huaiying Zhang (49:46):
Yep. It’s a feedback loop where the stress is there but then it’s the stress that allows it to survive.

Anirban Roychowdhury (49:51):
So that means in a stress condition, say in, under chemotherapy. So, cells which have these bodies may survive. Due to the heterogeneity, we know that it greatly affects the chemoresistance property of cancer. So, once chemotherapy is inducing these DNA damages or a stress condition, means it is selecting the cells which are having these particular bodies. So, is there any chance of that happening?

Huaiying Zhang (50:20):
Yeah. There’s some evidence believing that the stem cancer cells particularly like to use ALT pathway because they are so stressed, they allow them to survive. Also, I mentioned early on that when telomerease [inaudible 00:50:37] cancer cells under the stress condition under the telomerase inhibitors, some of them will not die and then adopt ALT pathway then to escape that inhibition. And this is actually a huge problem for the cancer therapy for telomerase inhibitors.

Anirban Roychowdhury (50:56):
Yes. The recent work, these chemotherapy drugs can partition into the condensates. So, it might be interesting that if these commonly used drugs it’s getting partitioned into these condensate bodies at the telomeres.

Huaiying Zhang (51:12):
That’s a great thought, yeah. I would imagine, like for the ones that are particularly enriched in the PML body would be very interesting to try with ALT cancers, right. But again, in general, right, targeting telomeres has this kind of drawback with it’s slow. So probably some sort of combination would be ideal where you cure rapidly with other drugs like [inaudible 00:51:37] drug. But then you inhibit the ALT pathway, so they don’t become like stem cancer cells. I think that would be very interesting.

Anirban Roychowdhury (51:51):
Thank you.

Mark Murcko (51:51):
That makes a lot of sense. Thank you. And so building some more on PML. Will had another couple of questions relating to that.

Will Chen (51:58):
Hey Huaiying.

Huaiying Zhang (52:00):
Hi.

Will Chen (52:00):
One question I had was… like Mark was saying that… So the reason I asked about whether there’s a stratification of the telomere mechanism is that in general DNA damage and chromosomal instability features in different kinds of cancer, subtypes of cancers. So this mechanism that you have of recruiting PML bodies, do you see any evidence or have you any ideas for whether that can be harnessed in other repair situations by the cancers?

Huaiying Zhang (52:30):
PML protein itself and PML bodies are implicated in DNA damage response. Particularly, I think, there are a lot of work early on showing PML body is a sensor for DNA damage. And there are also some work showing that PML protein form in as they call nanobody, that locally kind of form as a DNA damage response to help with the DNA damage response. I don’t think it’s clear how its involved. For example, in homologous recombination does it involve recruiting which protein? I don’t think that is clearly demonstrated, but there’s definitely a correlation or actually colocalization of PML nanobody to the damage [foci]. And there are more colocalization when cells actually fail to actually repair DNA damage. So you will see those kind of persistent DNA damage foci. Then goes to the PML body or I don’t know if the nuclear PML body or they actually just go to PML body because they move around and then they start to fuse. At least, I think it’s comfortable to say that the PML body is linked with the DNA damage response, but I don’t know if it’s a direct therapy.

Huaiying Zhang (54:00):
By just using [inaudible 00:54:01] or PML as body as a sensor for DNA damage then you could inhibit DNA damage response. For example, there are some drugs where you can inhibit DNA damage response and that if you also inhibit transcription and then that seems to be actually better for the therapy. So, they are, I think a sort of a combination, meaning because of the [cluster], because transcription and the DNA damage response and the PML body. So from that point of view, I can see that it might be reasonable. And also in general, PML body is a stress response, right. So for overexpression, for heat stress, for example even those kind of stresses PML body is involved in it. But again, what’s the particular role of PML body, I don’t think it’s super clearly defined as a molecular pathway.

Will Chen (54:58):
Excellent. Thank you.

Mark Murcko (54:59):
That’s a great answer. It’s a complicated topic. It’s not going to go away. That’s my prediction, is what we’ll keep learning more about this. So maybe we’re at a good stopping point. We’ve run a little bit over the hour, some great questions. Wonderful answers. Wonderful lecture.

Huaiying Zhang (55:15):
Thank you.

Mark Murcko (55:18):
This has been fantastic. Huaiying is an amazing speaker–clearly we’ve all learned that–and doing wonderful research. So thank you for your time and thank you for doing this for the whole community.

Huaiying Zhang (55:32):
Thank you for the opportunity.

Mark Murcko (55:34):
Sure. All right. Take care everyone. Thanks for joining us.

Huaiying Zhang (55:38):
Thanks. Bye.

Will Chen (55:39):
Thank you.

Mark Murcko (55:41):
Bye-bye.